MICROPLANKTON 209 



never achieved. Hensen himself had discovered that methods had to be varied to suit the study of 

 different organisms. 



Secondly, the necessity for wide, rapid coverage in collection, especially in the less known areas. 

 This often renders crude estimates of relative frequency of greater immediate value than limited series 

 of more accurate data, and was the main reason for our choice of an 'antiquated' method here. 

 The principle is epitomized in Gran's phrase 'a single "absolute" estimation of phytoplankton 

 would be about as valuable as a single temperature determination carried to the third decimal 

 place '. 



We have used Hensen's methods for subsampling and counting from routine vertical hauls of the 

 fine silk -J— m. net (N 50 V), but do not follow the early workers in regarding the hauls as repre- 

 senting a constant fraction of the entire contents of the theoretical volume of water filtered. The 

 estimations from counts we regard as roughly comparable indications of relative abundance, within 

 the limits of the very large differences involved in plotting (say) the estimated totals on a logarithmic 

 scale. They are, however, much more accurate as regards the relative importance of species within 

 single samples; and more accurate for small samples than for large ones that have to be diluted to 

 render subsamples small enough to count. 



We have found it easier to appreciate the relative frequencies involved by tabulating the actual 

 estimated numbers in primary tables, but in any form of graphic representation it becomes necessary 

 to use logarithmic or other functional ordinates. For the logarithmic reductions we have the classic 

 precedent of Professor Hentschel's work on the 'Meteor' centrifuge-plankton (Hentschel (1936)), and 

 the wide applications of logarithmic scales in dealing with many types of biological data have more 

 recently been well reviewed by Williams (1947). Other scale reductions are indicated on individual 

 diagrams, and it is hoped that sufficient tabular data are included to forestall queries arising from the 

 distortions unavoidable in graphic presentations of this kind. 



Recognizing the limitations of these numerical estimates, derived from roughly comparable hauls, 

 the necessity to adopt some arbitrary definition of dominance must still be faced; though we have 

 ventured to retain such useful expressions as 'frequent' and 'important' in their loose, generally 

 accepted sense. Now, the surveys were planned to include observations within two distinct types of 

 surface-water and the boundary region between them. Hence our criterion of dominance must be 

 applicable to individual samples. We cannot, for instance, employ some method like that of Sargent 

 and Walker (1948), who determined most satisfactory levels of significance (on a percentage basis) by 

 pooling results from a whole series of observations. Their method, one of the best we have yet seen, 

 demands, first, that the observations be restricted to a single water mass (though ' succession ' with 

 increase in 'age' of the surface-water may be apparent), and, secondly, adequate seasonal coverage; 

 so that ' succession ' should not be confused with ' sequence ' (i.e. changes in population due to invasion 

 of the area by a different water mass). 



Our definition derives direct from the raw data — the estimates of numbers of each category 

 recognized (some 12-50 categories) at each of the eighty-odd stations, arrayed in numerical order from 

 the original counts. We regard the first seven on each list as 'dominant'. By this means the local 

 importance of the more exclusively offshore species is not obliterated by the vast preponderance of 

 the inshore forms, as it would be if the results from the whole of this survey area were pooled. Further, 

 the local importance of some panthalassic species, present in the proportion of (say) 20,000/100,000 at 

 some impoverished offshore station, is not obscured by its presence in greater quantity — perhaps 

 200,000 — at a rich inshore station where the estimated total might be some 200 million, and the first 

 eight or ten categories all exceed two million. 



Considering only the thirty-nine stations from each survey that were repeated at approximately the 



